Simple model for clustering and ionic transport in ionomer membranes

H. L. Yeager and J. D. Malinsky. 1986,144-152. Abstract | PDF | PDF w/ Links. Article Options. PDF (534 KB) · Abstract · Citing Articles. Tools & Shar...
0 downloads 0 Views 544KB Size
1704

Macromolecules 1984, 17, 1704-1708

(16) Benoit, H.; Doty, P. J . Phys. Chem. 1953,57, 958. (17) Vink, H. Makromol. Chem. 1969, 122, 271. (18) Noda, I.; Tsuge, T.; Nagasawa, M. J . Phys. Chem. 1970, 74, 710. (19) Takahashi, A.; Nagasawa, M. J. Am. Chem. Soc. 1964,86.543. (20) Takahashi, A.; Kato, T.; Nagasawa, M. J . Phys. Chem. 1967, 71, 2001. (21) Brown, W.; Henley, D. Makromol. Chem. 1964, 79, 68. (22) Fisher, L. W.; Sochor,A. R.; Tan, J. S. Macromolecules 1977, 10, 949. (23) Meites, T.; Meites, L. Talanta 1972, 19, 1131. (24) Tricot, M.; Houssier, C. Macromolecules 1982, 15, 854. (25) Pierre, J. Ph.D. Thesis, LiBge, 1975. (26) Motowoka, M.; Norisuye, T.; Fujita, H. Polym. J . 1977,9,613. (27) Moore, W. R.; Brown, A. M. J . Colloid Sci. 1959, 14 (l),343. (28) Kay, P. J.; Treloar, F. E. Makromol. Chem. 1974, 175, 3207. (29) Hearst, J. E.; Stockmayer, W. J . Chem. Phys. 1962, 37, 1425.

(30) Hearst, J. E. J . Chem. Phys. 1963, 38, 1062. (31) Tricot, M.; Houssier, C.; Desreux, V.; van der Touw, F. Biophys. Chem. 1978,8, 221. (32) Tricot, M.; Houssier, C., to be published. (33) Yamakawa, H.; Yoshizaki, T. J . Chem. Phys. 1981, 75, 1016. (34) Rubingh, D. N.; Yu, H. Macromolecules 1976, 9, 681. (35) Schurr, J. M. Chem. Phys. 1980,45, 119. (36) Schurr, J. M.; Allison, S. A. Biopolymers 1981, 20, 251. (37) Manning, G. S. Biopolymers 1981, 20, 1751. (38) Kamide, K.; Saito, M. Eur. Polym. J. 1983, 19, 507. (39) Borochov, N.; Eisenberg, H.; Kam, Z. Biopolymers 1981,20, 231. (40) Hagerman, P. J. Biopolymers 1981, 20, 1503. (41) Hagerman, P. J. Biopolymers 1983, 22, 811. (42) Westra, S. W.; Leyte, J. C. Ber. Bunsenges. Phys. Chem. 1979, 83, 672. (43) Manning, G. S. Biophys. Chem. 1977, 7, 95.

Simple Model for Clustering and Ionic Transport in Ionomer Membranes Vandana K. Datye and Philip L. Taylor* Department of Physics, Case Western Reserve University, Cleveland, Ohio 44106

A. J. Hopfingert Department of Macromolecular Science, Case Western Reserve University, Cleveland, Ohio 44106. Received September 26, 1983

ABSTRACT: Cluster formation of the ionic side groups in an ionomer membrane of the Nafion type has been modeled and its relationship to ionic transport examined. Computer simulations show that the effect of electrostatic and elastic forces on the pendant ionic groups and their neutralizing counterions is such as to form a dipole layer at the surface of an ionic cluster. A model for selective transport is proposed based on the electrostatic potential created by the dipole layer. Qualitative agreement with experiment is obtained for the cluster size and current efficiency as a function of the equivalent weight of the membrane.

I. Introduction Ion-containing polymer membranes exhibit a sizable difference in permeability between oppositely charged ionic species. Although the origin of this phenomenon is not well understood, it is generally believed that permselectivity arises as a consequence of the electrostatic interaction at the ion exchange sites in the membrane.' The ionic groups of the ionomer phase separate into clusters, forming a hydrophilic ionic phase in a hydrophobic polymer matrix. In this paper we report the results of a modeling of cluster formation in an ionomer membrane and its relation to ionic transport through the membrane. In particular, we consider a system of the Nafion type and describe computer simulations of the spatial organization of ion dipoles within a cluster and have examined the relationship of this structure to ionic transport. The ionomer model that we have considered is highly simplified, containing only the most important ingredients in the problem, namely, the electrostatic and the elastic contributions to the energy. Our calculation is similar, in principle, to that of Eisenberg,2with the important difference that we do not assume any particular orientation for any dipole but rather treat these orientations as parameters that must be allowed to vary as the system minimizes its free energy and seeks its most favorable configuration. Perhaps the most interesting result to emerge from our simulations is that under the influence of electrostatic forces alone the ionic dipoles orient themselves to lie in the surface of the 'Alternate address: Department of Medicinal Chemistry, Searle Research and Development, Skokie, IL 60077.

spherical clusters. There is thus no effective dipole layer a t the cluster surface and the boundary is symmetric to the flow of cations and anions. However, as one will see in section 111, the process of cluster formation gives rise to an elastic energy that breaks this symmetry and leads to selective transport. 11. Definition of the Model While it is clear from experimental observation that clustering of the pendant-side-chain ionic groups does occur, the precise form of these clusters has not been determined. Some evidence exists3 to suggest that the clusters may deviate from spherical form, possibly even being tubular in nature. In the present work we will make the simpler assumption that the dipolar ionic groups are distributed over the surface of a spherical cluster. We will thus base our calculations on a geometry very similar to that of the inverted-micelle model proposed by Gierke.4 In choosing this model we are not denying the possibility that the dipoles may form a volume distribution within the cluster; we are rather taking as our first model the more plausible case in which the surface distribution of dipoles is not penetrated by segments of inert chain backbone, with the attendant energy cost. We further assume that the ionic groups, which are taken to be neutralized by counterions and exist in the form of ion pairs or dipoles, form a close-packed arrangement on the surface of the cluster, the nearest-neighbor distance being determined by the molecular pair potential. As mentioned in the Introduction, we consider only elastic and electrostatic effects. The driving force for clustering results from the fact that the electrostatic energy is lowered when dipoles

0024-9297/84/2217-1704$01.50/0 0 1984 American Chemical Society

Clustering in Ionomer Membranes 1705

Macromolecules, Vol. 17,No.9, 1984 aggregate. However, this stabilization is opposed by an increase in elastic energy on clustering that results from the conformational deformation of the polymer chains to which the fixed ions are attached. The equilibrium size of the cluster is determined by the balance of these opposing forces. We compute the electrostatic energy of the cluster by considering the dipole-dipole interaction between ion pairs. The electrostatic energy E,, per dipole is 1

(

= N(47I‘Kto)

X

g g --si.fij

i>j j=1

l?i

- ?j13

3 ~ ~ -4ij))(13i.(Fi 7 ~ - r.

I ?-~ ~

-1)))

Figure 1. Schematic illustration of the deformation of polymer chains due to clustering of the pendant ions.

(2.1)

~ 1 5

where f i i is the dipole moment of the ith dipole located at 7ion the surface of a sphere of radius r, eo is the permittivity of space, K is the dielectric constant, and N is the number of dipoles in the cluster. In our calculations we assume K = 4, i.e., the value for bulk poly(tetrafluoroethylene), which constitutes the polymer matrix in Ndion. The implications of this assumption are discussed in section V. We assume that the ionic dipoles are distributed uniformly over the surface of the spherical cluster. This uniform distribution is only exactly realizable for certain numbers of dipoles, a fact related to the limited number of regular polyhedra. Thus all dipoles are equivalent when located on a spherical surface at points corresponding to the vertices, centers of the faces, or the midpoints of the edges of a tetrahedron, cube, octahedron, dodecahedron, or icosahedron. This allows one to place 4,6,8,12,20,or 30 points on the spherical surface. Arrays of nearly equivalent points may be constructed by repeated centering of the faces of the various polyhedra, and in this way approximately uniform distributions of 42,80,120, and 162 points were constructed. Computer simulations were performed on clusters of dipoles to determine which orientations of the dipoles minimize the electrostatic energy. Initially a completely random distribution of orientations was constructed, and the dipoles allowed to relax into a self-consistently stable arrangement. This ground state is highly degenerate as a consequence of the symmetry of the array, and so different random initial conditions led to various final orientations. However, the fact that a unique final energy value was always attained showed that no false subsidiary minima were encountered and that the final orientations were all related to each other by simple rotations or reflections. In our simulations the effect of an external, applied field such as exists under membrane operating conditions is not included, because typically the applied field is several orders of magnitude smaller than the local dipole fields. To obtain an estimate of the elastic energy, we assume that the conformationaldeformation of the polymer chains connecting successive pendant ionic groups takes place in the following manner: Roughly half of the chains starting from the pendant ions terminate on the same cluster while the remaining chains terminate on a nearest-neighbor cluster (see Figure 1). The average end-bend separation for half of the chains is then simply the nearest-neighbor distance between dipoles on the spherical cluster, which we define to be do,and the average end-to-end separation (Ad) of the remaining chains is Ad = 2(R - r ) (2.2) where r is the radius of the cluster and R is the edge of

the cubic volume occupied by the N ions before clustering. Specifically

R3 = NWE/pNA

(2.3)

where p is the density of the membrane, NAis Avogadro’s number, and WE stands for the equivalent weight of the membrane (defined as the weight of acid polymer that will neutralize 1 equiv of base; i.e., WE roughly equals the monomer molecular weight). Using the simplest model of rubber elasticity5 and ignoring effects such as crystallinity in the sample, we find the elastic energy per dipole to be

where ( h2) is the mean square end-to-end chain length, T is the temperature, and kB is Boltzmann’s constant. Since do